Abstract
Breast milk samples from human immunodeficiency virus type 1 (HIV-1)-seropositive women were analyzed by polymerase chain reaction to determine the prevalence and determinants of HIV-1-infected cells in breast milk. Breast milk samples (212) were collected from 107 women, and 58% of the samples had detectable HIV-1 DNA. The proportion of HIV-1-infected cells in the milk samples ranged from 1 to 3255/104 cells. Breast milk samples with detectable HIV-1 DNA were more likely to be from women with absolute CD4 cell counts of <400 (odds ratio, 3.1; 95% confidence interval [CI], 1.5–7.0). Severe vitamin A deficiency (<20 μg/dL) was associated with a 20-fold increased risk of having HIV-1 DNA in breast milk among women with <400 CD4 cells/mm3 (95% CI, 2.1–188.5). Women with CD4 cell depletion, especially those with vitamin A deficiency, may be at increased risk of transmitting HIV-1 to their infants through breast milk.
Over the past few years, evidence has accumulated that breast milk plays an important role in mother-to-infant transmission of human immunodeficiency virus type 1 (HIV-1). Breast milk transmission of HIV-1 was clearly documented among women who developed primary HIV-1 infection in the postnatal period [1–3]. Among women with chronic HIV-1 infection (i.e., those seropositive at delivery), breast-feeding may double the rate of mother-to-child transmission of HIV-1 [4]. In the European Collaborative Study [4], the odds ratio (OR) for the association between breast-feeding and risk of infant infection was 2.25 (95% confidence interval [CI], 0.97–5.23). In a meta-analysis of data from 10 published studies, the estimated risk of mother-to-child transmission through breast milk was 29% (95% CI, 16%–42%) for postnatally infected women and 14% (95% CI, 7%–22%) for women already infected at delivery [5]. In an observational study in Nairobi, the mother-to-child transmission rate was 42%, and it was estimated that 32% of infant infections were attributable to prolonged breast-feeding (≥ 15 months) [6].
Breast milk protects infants against respiratory and intestinal infections and provides for nutritional development. These benefits are particularly important in developing countries, where safe alternatives to breast milk are often lacking and diarrheal and respiratory diseases commonly cause morbidity and mortality in children. Recognizing the benefits of breast-feeding and the potential hazards of alternative feeding in developing countries, the World Health Organization (WHO) recommends that HIV-seropositive women breast-feed their infants in settings where infant mortality rates due to infectious disease and malnutrition are high [7]. The importance of breast-feeding for children in developing countries must be balanced against the risks of breast-feeding by the large and growing number of HIV-infected women. Further understanding of the biology and determinants of breast milk transmission of HIV is needed to better counsel HIV-infected women.
In November 1992, we initiated an ongoing randomized clinical trial of breast-feeding versus formula-feeding in infants of HIV-seropositive women in Nairobi to determine the risk of breast milk transmission of HIV. A component of the clinical trial, which we present here, included analysis of breast milk for HIV-1 DNA to define the prevalence, concentration, and determinants (including maternal immunosuppression and vitamin A deficiency) of virus-infected cells in breast milk.
Methods
Study Population and Clinic Procedures
Pregnant women attending two Nairobi City Council antenatal clinics were tested for HIV. Seropositive women eligible for enrollment in the clinical trial responded to a standardized evaluation of demographic, social, sexual, and obstetric histories and were given physical examinations.
At 32 weeks of gestation, specimens were collected from the women for detection of Neisseria gonorrhoeae by culture, Chlamydia trachomatis by antigen detection (Clearview Chlamydia; Unipath, Nepean, Canada or Microtrak; Syva, San Jose, CA), and yeast by direct microscopy. Blood (15 mL) was drawn for differential and total white blood cell counts and enumeration of CD4 and CD8 lymphocytes using monoclonal antibodies (Becton Dickinson, San Jose) and flow cytometry (FACScan; Becton Dickinson). Serologic testing for syphilis was done using the rapid plasma reagin assay for screening and the Treponema pallidum hemagglutination assay for confirmation.
At the 32-week visit, women interested in participating in the clinical trial were randomized to either breast- or formula-feeding groups. Women received routine antenatal care in the research clinic. Within the first week after delivery, usually during the first 3 days, 5 mL of colostrum was manually expressed. At 6, 14, 26, 39, and 52 weeks, ≥ 10 mL of breast milk was obtained from all breast-feeding women.
Laboratory Methods
HIV serologic assays
Sera were tested for HIV antibodiesc using a syntheticp eptide ELISA (Behring, Ausgabe, Germany). Repeatedly reactive sera were confirmed using a different ELISA (Cambridge Biotech, Rockville, MD). Women testing positive on all three tests were considered to be HIV-1 seropositive.
Vitamin A assays
In serum or plasma obtained during the second or third trimester, vitamin A levels were measured using high-performance liquid chromatography. Vitamin A deficiency was defined as levels <30 μg/dL; severe vitamin A deficiency was defined as levels <20 μg/dL [8].
Processing of breast milk samples
Colostrum and breast milk samples were delivered to the laboratory with in 2 h of collection and centrifuged at 710 g for 20 min to yield three layers (lipids, supernatant, and cells). The lipid layer was removed using a wide-bore pipette and discarded. The supernatant was aliquoted and stored at −70°C. The cells were resuspended in 10 mL of RPMI 1640 and washed three times in RPMI by spinning at 300 g for 10 min, twice at room temperature and once at 4°C. The supernatant was discarded after this final wash, and the cells were suspended in 1 mL of RPMI freezing medium. The cell suspension was frozen at −70°C for 12 h before being transferred into liquid nitrogen for storage and transport to the University of Washington.
Detection of HIV DNA by polymerase chain reaction (PCR)
Breast milk cells were thawed at 37°C, counted, and pelleted by gentle centrifugation. Cells were resuspended in PBS and pelleted again. Cells (105–106) were lysed in 100 μL of lysis buffer (10 mM TRIS-HC1, pH 8.3, 50 mM KC1, 0.01% [wt/vol] gelatin, 0.45% NP-40, 0.45% Tween 20, and 0.6 mg/mL proteinase K), incubated at 56°C for 90 min, and boiled for 15 min to inactivate the proteinase K.
HIV-1 gag DNA sequences were detected by two rounds of nested PCR amplification of cell lysates, as previously described [9]. This PCR method has been shown to consistently amplify a single HIV-1 provirus among 104 uninfected cells [9].
All breast milk cell samples were tested in duplicate at 104 cells per PCR assay. Serial 5-fold dilutions of each positive sample were tested until HIV-gag DNA could no longer be detected. Using the last PCR-positive dilution, the number of HIV-infected cells per 104 cells was calculated.
Data Analysis
Data were analyzed using SPSS-PC (SPSS, Chicago), S-PLUS (MathSoft, Seattle), and Epi-Info (USD, Stone Mountain, GA) statistical programs. To assess correlates of HIV-1 DNA in colostrum, Student’s t and nonparametric median tests and analysis of variance were used to compare continuous variables; Pearson’s χ2 and Fisher’s exact tests were used for categorical variables. To estimate the strength of association between various correlates and presence or concentration of HIV-1 DNA in breast milk, ORs and 95% CIs were calculated.
Generalized estimating equations with logit link, binomial variance function, and exchangeable correlation structure were used to estimate the strength of correlation between various risk factors and presence of HIV-1 DNA in breast milk over time [10, 11]. In an initial model, the outcome was dichotomized (presence vs. absence of HIV-1 DNA in 104 breast milk cells) and univariate ORs were calculated. To evaluate a doseresponse relationship between correlates and concentration of HIV-l-infected cells in breast milk, a second analysis compared breast milk samples with low (≥10 infected cells/104) and high infected cells/104) virus concentrations with those with (I>10 no detectable HIV-1 DNA. To create dichotomous strata for continuous variables, cut points were determined by calculating the median value of each variable in the study population and rounding to a convenient number.
Results
Characteristics of the study population
One hundred seven women with a mean age of 24 years (range, 17–38) were included in the study. Seventy-one percent were married, and they reported a median of 3 lifetime sex partners.
HIV-related immunosuppression was assessed using clinical criteria and absolute CD4 lymphocyte counts. Using the modified WHO clinical case definition, none of the women had AIDS [12]. Of 105 women, 41 (39%) had HIV-related symptoms (fever or cough for >1 month, ≥ 10% loss of body weight, or history of skin rash or shingles); 6 (6%) of 100 women had clinical signs of HIV (rash, herpes zoster, thrush, or oral ulcers). Of 94 women with available absolute CD4 cell counts, 16% had <200 cells, 61% had 200–499, and 23% had ≥500. The mean CD4 cell count was 402 cells/mm3 (range, 46-850).
Among 97 women with measured levels of vitamin A, 17% had <20 μg/dL, 41% had 20–29.9, 24% had 30–39.9, and 19% had ≥40. Thus, 58% had vitamin A deficiency (<30 μg/dL) and 17% had severe deficiency (<20 μg/dL). The median level of vitamin A was 28.0 μg/dL (range, 10–58.7). None of the women took vitamin supplements either before or after delivery. Vitamin A levels were correlated with the percentage of CD4 cells (R = .15, P = .04), percentage of CD8 cells (R = −.15, P = .04), absolute CD8 cell count (R = −.15, P = .04), and CD4:CD8 ratio (R = .14, P = .05). Vitamin A levels were not correlated with absolute CD4 cell count (R = −.04, P = .6).
Prevalence and correlates of HIV-l-infected cells in breast milk
The 212 breast milk samples from 107 women included 77 obtained during the first week postpartum, 64 during days 8–90, 44 during months 3–6, 17 during months 6–9, and 10 >9 months after delivery. One hundred twenty-three (58%) of the samples had detectable HIV-1 DNA. As shown in figure 1, the prevalence of HIV-1 DNA was 51% in colostrum samples, 66% in samples obtained between 8 and 90 days, 64% in 3- to 6-month samples, 71% in 6- to 9-month samples, and 20% in samples obtained >9 months postpartum. The prevalence of HIV-1 DNA was significantly higher in samples obtained between 8 days and 9 months than in colostrum (P = .02) or samples obtained after 9 months (P = .01).
Figure 1.
Prevalence and concentration of HIV-1 DNA in breast milk cells. Percents indicate % with <10 or : 10 infected cells/104. mos, months.
A comparison of women with and without detectable HIV-1 DNA in colostrum is shown in table 1. The presence of HIV-1–infected cells in colostrum was not associated with maternal age, age at first sexual intercourse, lifetime number of sex partners, or past or current sexually transmitted diseases. Women with HIV-1-infected cells in colostrum had lower mean CD4:CD8 ratios (P = .07) and more often had HIV-related symptoms (P = .05) and a higher percentage of CD8 cells (P = .04).
Table 1.
Association between HIV-1 DNA in colostrum and maternal characteristics.
| Characteristic | PCR negative (n = 38) |
PCR positive (n = 39) |
P |
|---|---|---|---|
| Demographic variables | |||
| Age (years) | 23 | 25 | .2 |
| Parity | 1 | 1 | .4 |
| Married | 28/38 (74) | 29/39 (74) | .9 |
| Sex history | |||
| Age at first intercourse (years) | 17 | 17 | .8 |
| Lifetime no. of sex partners* | 3 | 3 | .7 |
| No. of sex partners during pregnancy* | 1 | 1 | .2 |
| Sexually transmitted diseases | |||
| Past | 15/38 (40) | 11/39 (28) | .3 |
| Current | |||
| Syphilis | 0/36 (0) | 1/38 (3) | 1.0 |
| Gonorrhea | 3/35 (9) | 2/36 (6) | .7 |
| Chlamydia | 6/34 (18) | 2/34 (6) | .3 |
| Genital infections | |||
| Candidiasis | 10/34 (29) | 11/37 (30) | 1.0 |
| Genital warts | 5/38 (13) | 6/38 (16) | .7 |
| Disease status | |||
| HIV-related symptoms† | 12/38 (32) | 21/39 (54) | .05 |
| HIV-related signs‡ | 1/38 (3) | 3/38 (8) | .6 |
| T cell indices | |||
| Absolute CD4 cell count | 424 | 367 | .2 |
| Percent CD4 cells | 25 | 23 | .4 |
| Absolute CD8 cell count† | 713 | 700 | .7 |
| Percent CD8 cells | 45 | 50 | .04 |
| CD4:CD8 ratio | 0.6 | 0.5 | .07 |
| Total lymphocyte count | 1702 | 1579 | .4 |
| Vitamin A (μg/dL)* | 30.8 | 28.0 | .2 |
| μWomen with absolute CD4 cells <400* | 32.9 | 24.2 | .07 |
| μWomen with absolute CD4 cells ≥00* | 28.7 | 29.5 | 1.0 |
NOTE. Unless otherwise stated, data are means or no./no. total (%). Subjects were negative or positive for HIV-1, as assessed by polymerase chain reaction (PCR).
Median.
Fever or cough for >1 month, ≥10% weight loss, or history of rash or shingles.
Rash, herpes zoster, thrush, or oral ulcers.
Overall, median vitamin A levels did not differ between women with or without detectable HIV-1 DNA in colostrum. However, in a stratified analysis, among women with <400 CD4 cells/mm3, there was a trend for a relationship between colostral HIV-1 DNA and lower vitamin A levels (24.2 vs. 32.9 μg/dL, P = .07). The prevalence of colostral HIV-1 DNA was determined for four strata of vitamin A levels. There was a monotonic increase in the prevalence of HIV-1-infected cells in colostrum with declining maternal levels of vitamin A: 33% at ≥40 μg/dL, 50% at 30–39.9 μg/dL, 54% at 20–29.9 μg/dL, and 67% at <20 μg/dL. After being stratified by CD4 cell count, this inverse relationship between colostral HIV-1 DNA and vitamin A levels was statistically significant for women with <400 CD4 cells/mm3 (P = .01; figure 2) but not for women with higher CD4 cell counts.
Figure 2.
Association of maternal vitamin A levels with prevalence of HIV-1 DNA in colostrum for women with absolute CD4 cells <400. OR, odds ratio; PCR, polymerase chain reaction.
Maternal correlates of HIV-1 DNA in breast milk samples obtained at any point during lactation are shown in table 2. After an adjustment for time since delivery, the presence of HIV-1-infected cells in breast milk was significantly correlated with <400 CD4 cells/mm3 (OR, 3.1; 95% CI, 1.5–7.0), <23% CD4 cells (OR, 2.4; 95% CI, 1.1–5.2), >47% CD8 cells (OR, 3.2; 95% CI, 1.5–6.9), and CD4:CD8 ratio <0.5 (OR, 2.8; 95% CI, 1.3–5.9). In this analysis, the presence of HIV-1–related signs or symptoms was not significantly associated with HIV-l–infected breast milk cells. Vitamin A deficiency was unrelated to the presence of HIV-1 DNA in breast milk among the total sample of women. However, among women with <400 CD4 cells/mm3, there was a trend for a relationship between breast milk provirus and vitamin A levels <30 μg/dL (OR, 3.1; 95% CI, 0.8-11.5). This association is examined in more detail in table 3. As with the above analysis for colostrum, severe vitamin A deficiency (<20 μg/dL) was associated with a significantly increased risk of HIV-1-infected cells in breast milk (OR, 19.7; 95% CI, 2.1–188.5) among women with CD4 cell depletion.
Table 2.
Risk factors for HIV-1 DNA in breast milk throughout follow-up.
| Risk factor | Any HIV-1–infected cells |
P | Low virus load | P | High virus load | P |
|---|---|---|---|---|---|---|
| Age >23 years | 1.6 (0.8–3.4) | .2 | 1.5 (0.7–3.2) | .3 | 1.8 (0.8–4.4) | .2 |
| Age at first intercourse <17 years | 1.6 (0.7–3.3) | .3 | 1.5 (0.7–3.4) | .3 | 1.6 (0.7–3.9) | .3 |
| >3 lifetime sex partners | 2.0 (0.9–4.5) | .08 | 1.8 (0.8–4.3) | .2 | 2.2 (0.8–5.7) | .1 |
| HIV-related symptoms* | 1.4 (0.7–2.8) | .4 | 1.4 (0.6–3.2) | .4 | 1.1 (0.5–2.7) | .8 |
| HIV-related signs† | 5.0 (0.6–42.5) | .1 | 3.8 (0.4–33.9) | .2 | 7.9 (0.8–74.5) | .07 |
| Absolute CD4 cell count <400 cells | 3.1 (1.5–7.0) | .004 | 2.3 (1.0–5.2) | .04 | 4.2 (1.6–11.2) | .004 |
| Percent CD4 cells <23 | 2.4 (1.1–5.2) | .02 | 2.1 (0.9–5.0) | .09 | 3.4 (1.4–8.6) | .009 |
| Absolute CD8 cell count >800 cells | 1.5 (0.7–3.2) | .3 | 1.8 (0.8–4.0) | .1 | 1.4 (0.6–3.3) | .5 |
| Percent CD8 cells >47 | 3.2 (1.5–6.9) | .002 | 3.2 (1.4–7.5) | .006 | 3.5 (1.4–8.7) | .007 |
| CD4:CD8 ratio <0.5 | 2.8 (1.3–5.9) | .008 | 2.6 (1.1–6.2) | .03 | 3.6 (1.5–8.8) | .005 |
| Lymphocyte count <1700 cells/mm3 | 1.4 (0.7–2.9) | .4 | 1.1 (0.5–2.5) | .7 | 1.5 (0.6–3.6) | .4 |
| Vitamin A <30 μig/dL | ||||||
| Women with absolute CD4 cells <400 | 3.1 (0.8–11.5) | .1 | 2.6 (0.6–11.0) | .2 | 3.7 (0.8–16.0) | .08 |
| Women with absolute CD4 cells >400 | 1.2 (0.4-3.4) | .8 | 1.6 (0.5-5.2) | .5 | 0.6 (0.1-2.3) | .4 |
NOTE. Data are odds ratios (95% confidence intervals) adjusted for time since delivery. For each column, comparison group was breast milk samples with no detectable HIV-l-infected cells. Low and high virus load = <10 and ≥10, respectively, HIV-l–infected cells/104 cells.
Fever or cough for >1 month, ≥10% weight loss, or history of rash or shingles.
Rash, herpes zoster, thrush, or oral ulcers.
Table 3.
Association of maternal serum or plasma levels of vitamin tion of HIV-1 DNA in breast milk.
| Level of vitamin A (μg/dL) | Any HIV-1– infected cells |
P | Low virus load | P | High virus load | P |
|---|---|---|---|---|---|---|
| Women with absolute CD4 cells <400 | ||||||
| <20 | 19.7 (2.1–188.5) | .01 | 17.8 (1.6–202.4) | .02 | 21.3 (1.9–239.1) | .01 |
| 20–29.9 | 2.7 (0.4–18.3) | .3 | 2.8 (0.3–25.9) | .3 | 2.7 (0.3–23.8) | .4 |
| 30–39.9 | 3.5 (0.7–18.6) | .15 | 3.9 (0.6–26.6) | .2 | 3.0 (0.4–21.5) | .3 |
| ≥40 | 1.0 | 1.0 | 1.0 | |||
| Women with absolute CD4 cells ≥400 | ||||||
| <20 | 0.4 (0.1–3.0) | .4 | 0.4 (0.0–5.7) | .5 | 0.2 (0.0–2.6) | .2 |
| 20–29.9 | 2.0 (0.7–5.6) | .2 | 3.9 (1.0–15.6) | .06 | 0.7 (0.2–2.4) | .5 |
| 30–39.9 | 1.6 (0.3–8.2) | .6 | 2.6 (0.4–18.3) | .3 | 0.8 (0.1–6.2) | .8 |
| ≥40 | 1.0 | 1.0 | 1.0 |
NOTE. Data are odds ratios and 95% confidence intervals adjusted for time since delivery. For each column, comparison group was breast milk samples with no detectable HIV-l–infected cells. Low and high virus load = <10 and ≥10 HIV-l–infected cells/104 cells, respectively.
During follow-up, abnormalities of the breast (i.e., nipple cracking or bleeding or breast abscess or mastitis) were present on physical examination in 3 of 66 women during 4 of 135 visits. There was no correlation between presence of breast abnormality and HIV-1 DNA detection in breast milk.
Concentration of HIV-1–infected cells and correlates of high virus titer
Because the PCR assays were done using 104 breast milk cells in the undiluted sample, the lower limit of detection of our assay was 1 infected cell/104 breast milk cells. The proportion of HIV-1–infected cells among positive samples ranged from 1 to 3255/104 cells, a difference of >3 logs. The median concentration of HIV-1–infected cells/104 breast milk cells was 2.5 in colostrum, 8.2 in samples obtained 8-90 days postpartum, 3.5 at 3–6 months, 2.5 at 6–9 months, and 1.4 at >9 months.
Breast milk samples were classified into three strata: no detectable virus, low virus load (<10 infected cells/104 cells), and high virus load (≥10 infected cells/104 cells). As shown in figure 1, the prevalence of high virus load was greatest at 8–90 days (31% vs. 20% at 0–7 days, P = .08) and then declined to zero after 9 months.
Using generalized estimating equations, we compared breast milk samples with low and high virus loads with samples with no detectable viral DNA to determine correlates of increased HIV-1 concentration in breast milk. The ORs for the associations between HIV-1 DNA and four measures of immunosuppression (absolute CD4 cell count <400, <23% CD4 cells, >47% CD8 cells, and CD4:CD8 ratio <0.5) increased from low to high virus load groups (table 2). For example, the OR was 2.3 (95% CI, 1.0–5.2) for the association between absolute CD4 cell count <400 and low virus concentration in breast milk and it was 4.2 (95% CI, 1.6–11.2) for high virus concentration.
Discussion
This study has yielded four important findings. First, HIV-1–infected cells were detectable in most (58%) breast milk samples, and this prevalence remained high from birth to 9 months. In a previously published study, the prevalence of HIV DNA among 47 Haitian women was 70% in colostrum and 47% in breast milk obtained 3–12 months postpartum [13]. Van de Perre, et al. [14] reported a prevalence of HIV DNA of 47% in breast milk samples obtained 15 days postpartum and 20% in 6-month samples among Rwandan women. In a small study from Thailand, HIV-1 DNA was detected in 44% of colostrum or transitional milk samples [15]. In contrast to findings in previous studies, we found a significantly higher prevalence of HIV-1 DNA in mature milk than in colostrum.
Second, our study provides the first quantitative determinations of the concentration of HIV-1 DNA in breast milk cells. There was a wide range of proportions of HIV-1 -infected cells per total number of cells (>3 log difference), and in the breast milk sample with the highest concentration, 33% of breast milk cells were infected. For comparison, this was considerably higher than the maximal prevalence of 1% infection of peripheral blood mononuclear cells we detected among women participating in our clinical trial.
The very high concentration of infected cells in some breast milk samples suggests that infants may be exposed to a significant oral dose of virus-infected cells. At any time point, the degree of such exposure would be a function of the proportion of cells infected with HIV-1, the concentration of cells per volume of milk, and the volume of milk ingested. The cellular concentration of colostrum is 10- to 100-fold that of mature milk [16, 17], while the volume of intake after the first few days of life increases by 5- to 10-fold [18]. Our data permit an estimation of the daily dose of HIV-1–infected cells that would be ingested by an infant breast-feeding at different time points during lactation from an infected mother who was shedding virus in breast milk. During the first few days of life, an infant who ingests 100 mL/day of breast milk, with a cellular concentration of 106 cells/mL [16], would be exposed to 25,000 HIV-l-infected cells per day (assuming 2.5 HIV-l-infected cells/104 cells). Although it has been proposed that withholding of colostrum and institution of breastfeeding after the first week of life might be evaluated as a possible intervention to reduce postnatal transmission of HIV-1 (as has been successfully done with the lentivirus caprine arthritis encephalitis virus in goats [19]), our calculations suggest that the level of provirus exposure remains high. At 6 months, for example, an infant who drinks 800 mL of breast milk/day [18] would ingest 24,000 HIV-l-infected cells/day (assuming a cellular concentration of 105 cells/mL [16] and a concentration of HIV-1 -infected cells of 3.0/104 cells).
These estimated rates must be interpreted with caution. The detection of integrated cellular HIV-1 DNA does not necessarily imply presence of infectious virus. Unfortunately, culture detection of replicating HIV-1 virus in breast milk remains a challenge because of inhibitory factors in human milk [20, 21]. Nonetheless, other investigators have found a significant correlation between detection of HIV-1 DNA in breast milk and risk of mother-to-child transmission, suggesting that the detection of provirus may be a reasonable surrogate for detection of virus per se [14].
The third finding we have demonstrated is a strong correlation between maternal immunosuppression and prevalence and quantity of breast milk HIV-1 DNA. Women with prenatal absolute CD4 cell counts of <400 had a 3.1-fold increased likelihood of shedding HIV-1 provirus in breast milk and a 4.2-fold increased likelihood of high provirus concentration in breast milk. This association is analogous to the increased frequency of plasma viremia and higher virus concentration in peripheral blood mononuclear cells seen with advanced HIV disease [22]. Our results are consistent with those of Van de Perre, et al. [14], who found an inverse association between the presence of HIV-1 DNA in breast milk samples obtained 15 days postpartum and maternal CD4:CD8 ratio. In contrast, Ruff, et al. [13] found no correlation between maternal immune status (measured by percent CD4 cells) and HIV-1 DNA in breast milk, but the power of the study was limited by small sample size.
Fourth, we found a significant inverse linear correlation between prenatal vitamin A levels and prevalence of HIV-1–infected cells in colostrum among women with <400 CD4 cells but not among women with higher CD4 cell counts. A recent study in Malawi found an inverse relationship between maternal vitamin A levels and risk of mother-to-child transmission of HIV-1 [23]. Women with vitamin A deficiency had a 3- to 4-fold increased risk of transmitting HIV-1 to their infants. Vitamin A has important immune regulatory functions as well as an effect on mucosal epithelial integrity, as recently reviewed by Semba [8]. It is biologically plausible that the humoral and cellular immune impairment associated with low levels of vitamin A could result in the increased prevalence of HIV-l -infected cells in breast milk that we observed. The fact that there was effect modification by the level of CD4 cells was unexpected, but it is possible that the impact of vitamin A deficiency on immune function may be more pronounced in the setting of preexisting immune impairment. If vitamin A truly affects the presence and concentration of HIV-1 in breast milk and the likelihood of vertical transmission, the fact that >50% of the pregnant women in our cohort were vitamin A deficient (17% were severely deficient) suggests that replenishment of vitamin A to normal physiologic levels could potentially have a significant impact in this kind of population.
What are the implications of these results? As outlined in the introduction, there is compelling evidence that breast-feeding contributes substantially to vertical transmission of HIV-1. Our results suggest that most HIV-1-seropositive women shed HIV-1–infected cells in breast milk, that the concentration of HIV-1–infected cells is sometimes very high, and that a breast-feeding infant may ingest tens or hundreds of thousands of HIV-l-infected cells per day. Women with advanced HIV disease and vitamin A deficiency may be at particular risk of transmitting HIV-1 through breast milk. Better understanding of the level and timing of risk for HIV-1 transmission through breast milk is urgently required to inform policy on appropriate infant feeding counseling for HIV-infected women, particularly in areas and populations where formula-feeding is associated with considerable morbidity. In addition, our data suggest that the evaluation of vitamin A supplementation in pregnant and breast-feeding women through randomized clinical trials is warranted.
Acknowledgments
We thank the clinical and laboratory research staff who made this study possible; the Nairobi City Council for permission to use the Umoja and Jericho clinics; the administration of Kenyatta National Hospital for permission to use various clinical facilities; Hjordis Foy for review of the manuscript; and Susan Mello for assistance with manuscript preparation.
Financial support: National Institutes of Health (HD-23412, T22-TW00001, D43-TW00007, and Clinical Nutrition Research Unit, DK-35816).
Footnotes
Informed consent was obtained from all patients. This study was reviewed and approved by ethical review committees of the University of Washington and of Nairobi. Human experimentation guidelines of the US Department of Health and Human Services were followed in the conduct of research.
References
- 1.Ziegler JB, Cooper DA, Johnson RO, Gold J. Postnatal transmission of AIDS-associated retrovirus from mother to infant. Lancet. 1985;1:896–8. doi: 10.1016/s0140-6736(85)91673-3. [DOI] [PubMed] [Google Scholar]
- 2.Palasanthiran P, Ziegler JB, Stewart GJ, et al. Breast-feeding during primary maternal human immunodeficiency virus infection and risk of transmission from mother to infant. J Infect Dis. 1993;167:441–4. doi: 10.1093/infdis/167.2.441. [DOI] [PubMed] [Google Scholar]
- 3.Van de Perre P, Simonon A, Msellati P, et al. Postnatal transmission of human immunodeficiency virus type 1 from mother to infant: a prospective cohort study in Kigali, Rwanda. N Engl J Med. 1991;325:593–8. doi: 10.1056/NEJM199108293250901. [DOI] [PubMed] [Google Scholar]
- 4.European Collaborative Study Risk factors for mother-to-child transmission of HIV-1. Lancet. 1992;339:1007–12. doi: 10.1016/0140-6736(92)90534-a. [DOI] [PubMed] [Google Scholar]
- 5.Dunn DT, Newell ML, Ades AE, Peckham CS. Risk of human immunodeficiency virus type 1 transmission through breastfeeding. Lancet. 1992;340:585–8. doi: 10.1016/0140-6736(92)92115-v. [DOI] [PubMed] [Google Scholar]
- 6.Datta P, Embree JE, Kreiss JK, et al. Mother-to-child transmission of human immunodeficiency virus type 1: report from the Nairobi study. J Infect Dis. 1994;170:1134–40. doi: 10.1093/infdis/170.5.1134. [DOI] [PubMed] [Google Scholar]
- 7.Global Programme on AIDS Consensus statement from the WHO/UNI-CEF consultation on HIV transmission and breast feeding. Wkly Epidemiol Rec. 1992;7:177–84. [PubMed] [Google Scholar]
- 8.Semba RD. Vitamin A, immunity, and infection. Clin Infect Dis. 1994;19:489–99. doi: 10.1093/clinids/19.3.489. [DOI] [PubMed] [Google Scholar]
- 9.Moss GB, Overbaugh J, Welch M, et al. Human immunodeficiency virus DNA in urethral secretions in men: association with gonococcal urethritis and CD4 cell depletion. J Infect Dis. 1995;172:1469–74. doi: 10.1093/infdis/172.6.1469. [DOI] [PubMed] [Google Scholar]
- 10.Zeger SL, Liang K. Longitudinal data analysis for discrete and continuous outcomes. Biometrics. 1986;42:121–30. [PubMed] [Google Scholar]
- 11.Zeger SL, Liang K, Albert PS. Model for longitudinal data: a generalized estimating equation approach. Biometrics. 1988;44:1049–60. [PubMed] [Google Scholar]
- 12.De Cock KM, Selik RM, Soro B, Gayle H, Colebunders RL. AIDS surveillance in Africa: a reappraisal of case definitions. Br Med J. 1991;303:1185–8. doi: 10.1136/bmj.303.6811.1185. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Ruff AJ, Coberly J, Halsey NA, et al. Prevalence of HIV-1 DNA and p24 antigen in breast milk and correlation with maternal factors. J Acquir Immune Defic Syndr. 1994;7:68–73. [PubMed] [Google Scholar]
- 14.Van de Perre P, Simonon A, Hitimana D, et al. Infective and anti-infective properties of breastmilk from HIV-l-infected women. Lancet. 1993;341:914–8. doi: 10.1016/0140-6736(93)91210-d. [DOI] [PubMed] [Google Scholar]
- 15.Buranasin P, Kunakorn M, Petchclai B, et al. Detection of human immunodeficiency virus type 1 (HIV-1) proviral DNA in breast milk and colostrum of seropositive mothers. J Med Assoc Thai. 1993;76:41–4. [PubMed] [Google Scholar]
- 16.Ogra SS, Ogra PL. Immunologic aspects of human colostrum and milk. II. Characteristics of lymphocyte reactivity and distribution of E-rosette forming cells at different times after the onset of lactation. J Pediatr. 1978;92:550–5. doi: 10.1016/s0022-3476(78)80286-8. [DOI] [PubMed] [Google Scholar]
- 17.Goldman AS, Garza C, Nichols BL, Goldblum RM. Immunologic factors in human milk during the first year of lactation. J Pediatr. 1982;100:563–7. doi: 10.1016/s0022-3476(82)80753-1. [DOI] [PubMed] [Google Scholar]
- 18.Riordan J. The biologic specificity of breastmilk. In: Riordan J, Auerbach KG, editors. Breastfeeding and human lactation. Jones and Bartlett; Boston: 1993. p. 107. [Google Scholar]
- 19.P’eretz G, Bugnard F, Calavas D. Study of a prevention programme for caprine arthritis-encephalitis. Vet Res. 1994;25:322–6. [PubMed] [Google Scholar]
- 20.Newburg DS, Viscidi RP, Ruff A, Yolken RH. A human milk factor inhibits binding of human immunodeficiency virus to the CD4 receptor. Pediatr Res. 1992;31:22–8. doi: 10.1203/00006450-199201000-00004. [DOI] [PubMed] [Google Scholar]
- 21.Orloff SL, Wallingford JC, McDougal JS. Inactivation of human immunodeficiency virus type 1 in human milk: effects of intrinsic factors in human milk and of pasteurization. J Hum Lactation. 1993;9:13–7. doi: 10.1177/089033449300900125. [DOI] [PubMed] [Google Scholar]
- 22.Ho DD, Moudgil T, Alam M. Quantitation of human immunodeficiency virus type 1 in the blood of infected persons. N Engl J Med. 1989;321:1621–5. doi: 10.1056/NEJM198912143212401. [DOI] [PubMed] [Google Scholar]
- 23.Semba RD, Miotti PG, Chiphangwi JD, et al. Maternal vitamin A deficiency and mother-to-child transmission of HIV-1. Lancet. 1994;343:1593–7. doi: 10.1016/s0140-6736(94)93056-2. [DOI] [PubMed] [Google Scholar]